[LEAPSECS] joint BIPM/ITU meeting

[Ian Batten]

On 14 Sep 2013, at 16:20, "Poul-Henning Kamp" <phk at phk.freebsd.dk> wrote:


> In message <D109BC3D-680E-4514-B500-14721F96441D at batten.eu.org>, Ian Batten wri

> tes:

>> 

>> On 13 Sep 2013, at 14:57, Steve Allen <sla at ucolick.org> wrote:

>> 

>>> In stark contrast to the usual ITU-R pattern and the previous workshop

>>> held by BIPM at the Royal Society, the presentations for next week's

>>> workshop in Geneva are being published

>>> 

>>> http://www.itu.int/oth/R0A0E000096/en

>>> 

>>> As Mike Meyers used to say "Talk amongst yourselves".

>> 

>> The Japanese presentation is genuinely deranged. 

> 

> And you totally missed, purely by accident I pressume, the bit about

> leap-seconds happening during business-days there, right ?


Of course, n a country in which the timezone is UTC+9, it will be 0900 on July 1 when a leapsecond set for 23:59:60 June 30 fires.  That makes an outage starting 0925 and finishing 2241 (both JST) on June 30 even more fascinating as a case of an operating taking different codepaths for an impending leap second.   What kernels even have an interface to accept this information?  What codepaths will be different 23:35 before a leap second?  

Perhaps the presentation as delivered will be more convincing, but just pointing to system outage in a 48 hour window either side of a leapsecond and say "look!  it was the leapseconds that did it!" is not hugely convincing.

ian